Mutational Landscape of RNA Binding Proteins in Human Cancers

Abstract

RNA Binding proteins (RBPs) are a class of regulatory molecules pivotal in orchestrating several post-transcriptional mechanisms. Recent studies have shown a significant number of them to be up-regulated in various cancer genomes and few of them were also shown to be contributing directly to cancer progression (1). Although variations in the gene expression levels of these proteins across cancers have been known, the cause for such alteration remains to be understood. A plausible hypothesis for this aberrant expression of RNA Binding proteins could be attributed to the occurrence of atypical mutations in the genes encoding for these proteins. In the present study, we aim to delineate the mutational landscape of ~1300 RNA Binding proteins (2) in ~6000 cancer genomes from The Cancer Genome Atlas (TCGA). Our analysis revealed that RBPs have an average of ~3 mutations per Mb across 26 tumor types. We identified ~600 RBPs to be significantly mutated in at least one tumor type (Corrected p<0.01, Fisher's exact test) suggesting that RBPs are as likely to be mutated as NonRBPs (p=0.9, Fisher's exact test). Among these, genes encoding for KMT2C, AHNAK and PLEC were found to be significantly mutated in at least 9 cancers suggesting common players of regulation across different tumor types. Furthermore, a comparison of different mutation types among significantly and non-significantly mutated RBPs revealed that the former genes are highly enriched for mutation types like inframe deletion (p < 8E-06, Wilcox test) and missense mutation (p < 3E-05, Wilcox test). Further, we identified RBPs (driver genes) whose function could be impaired due to the nonsynonymous SNPs, using OncodriveFM framework. We identify ~200 RBPs which can be defined as drivers in at least one tumor type (Corrected p<0.01). Of these, AHNAK was found to be mutated in at least 11 tumor types suggesting a common regulatory pathway affected in majority of the tumor types. Further analysis of the RNA levels of the driver genes between the patient groups with and without these deleterious mutations revealed ~30 RBPs (15% of the drivers) to be significantly altered in their expression levels between these groups (p < 0.05, Wilcox test). Protein-protein interaction network analysis of the driver genes identified in at least two tumor types revealed the presence of a cluster of mutated proteins involved in the spliceosomal machinery suggesting a plausible mechanism for tumorigenesis.